Recently, major media outlets have published alarmist headlines about a new study linking butter consumption with increased mortality and cancer risk. These headlines suggest a direct causal relationship that, as we'll see, is far from being demonstrated by the evidence presented. Let's analyze this research in depth to understand its true implications.
The misleading headlines
Major media outlets have presented categorical conclusions based on this research:
- Newsday: "Butter may shorten your life and increase cancer risk"
- Fortune Magazine: "A simple dietary change can help you fight cancer and heart disease"
- NBC News: "Eating less butter can improve health and protect against heart disease and cancer"
These headlines suggest a definitive cause-and-effect relationship, something no observational study can establish. This sensationalist presentation significantly distorts the research's inherent limitations.
Detailed study analysis
The study was published in JAMA Internal Medicine, a respected journal of the American Medical Association, under the title "Butter and vegetable oil intake and mortality." Let's examine its key methodological characteristics:
1. Study design: observational and its limitations
This is a prospective observational cohort study, which means:
- Cannot establish causality, only associations
- No randomization of participants to different groups
- Confounding variables are difficult to control completely
Basic epidemiology establishes that these studies serve to generate hypotheses, not to confirm them. Headlines claiming causal relationships based on this data commit a fundamental error.
2. Data collection through questionnaires
The study used food frequency questionnaires every four years, which presents significant problems:
- Memory bias: People imperfectly remember their food consumption
- Cumulative averages: Don't accurately reflect changing eating habits
- No objective verification: Absence of biological markers to verify reported consumption
These questionnaires are known in nutritional research for their imprecision and tendency to underestimate or overestimate consumption.
3. Inadequately controlled confounding variables
Although researchers attempted to adjust for variables like BMI, smoking, alcohol, and physical activity, fundamental problems persist:
- Healthy user bias: People who avoid butter also tend to adopt other healthy behaviors
- Socioeconomic factors: Not completely controlled but significantly influence outcomes
- Unrecorded dietary changes: Participants modify their diet for health reasons not captured
These confounding factors cannot be eliminated simply through statistical adjustments.
4. Problematic grouping of vegetable oils
The study compared butter with grouped "vegetable oils," including:
- Olive oil
- Canola oil
- Soybean oil
- Corn oil
- Vegetable shortening
This heterogeneous grouping combines oils with very different nutritional profiles, from minimally processed extra virgin olive oil to highly processed industrial oils.
5. Absence of control over processing and cooking
The study did not differentiate between:
- Raw oils vs. oils cooked at high temperatures
- Industrial processing (refined, bleached, deodorized)
- Formation of pro-inflammatory compounds during cooking
Oxidation and toxic compounds formed during heating can significantly influence health effects.
6. Death cause classification
Causes of death were determined through death certificates, a notoriously imprecise method:
- Doctors frequently estimate or assume causes when there's no clear evidence
- Autopsies are not performed in most cases
- Death certificates are often completed with limited information
As a physician with over two decades of experience, I can confirm that completing death certificates often involves educated guesses rather than scientific certainties.
Interpretation of statistical results
Marginal risk ratios
The study reported hazard ratios of:
- Butter: 1.15 (confidence interval not detailed)
- Vegetable oils: 0.84 (confidence interval not detailed)
These values represent statistically weak associations that barely exceed the significance threshold. In epidemiology, values below 2.0 are generally considered weak associations susceptible to bias.
Perspective on effect magnitude
For context, hazard ratios for:
- Smoking and lung cancer: >20.0
- Severe obesity and type 2 diabetes: >5.0
Weak associations like those found in this study frequently disappear in more rigorous or controlled research.
Institutional context of the research
Affiliations and funding
Some authors are affiliated with institutions that have historically promoted plant-based diets, like Harvard's Chan School of Public Health.
This institutional context can influence:
- Study design
- Result interpretation
- Emphasis on certain conclusions
Without randomization or blinding, researcher biases can subtly influence multiple aspects of the study.
Practical implications: what does it really mean?
For people on low-carbohydrate diets
If you follow a ketogenic, carnivore, or low-carbohydrate diet:
- This study provides no convincing evidence to eliminate butter
- Your personal results (blood markers, body composition, general well-being) have greater relevance for your individual case
- Observational studies have fundamental limitations in their applicability to specific individuals
For nutritional science
This study illustrates systemic problems in nutritional research:
- Excessive reliance on observational studies
- Media amplification of preliminary results
- Lack of context in scientific communication
- Influence of established nutritional paradigms
Conclusion: a balanced perspective
Alarmist headlines about butter represent an excessive interpretation of limited observational data. Fundamental methodological problems include:
- Inherent inability to establish causality
- Imprecise data collection methods
- Residual confounding by multiple factors
- Problematic grouping of diverse oils
- Imprecise outcome measurement
Instead of reacting to sensationalist headlines, consider:
- Evaluating your individual response to different fats through regular blood tests
- Using personal results (energy, body composition, inflammatory markers) as a guide
- Maintaining healthy skepticism about simplistic nutrition claims
Nutritional science requires controlled, randomized, and mechanistic studies to provide definitive answers. Meanwhile, observable personal results remain the most reliable guide for individual dietary decisions.